Skip to content

Repository files navigation

Haired

A free, open-source stealth pair-programming and interview AI assistant for macOS and Windows.

Haired answer overlay on a technical workspace

Download · Website · Features · Providers

What is Haired?

Haired sits quietly on your desktop until you need it. Select any permitted screen region and get a streamed coding answer in an always-on-top overlay—no context switching, no browser tab, no account required.

  • Stealth pair programming — ask about code, errors, tests, or diagrams without leaving your editor.
  • Interview Mode — structured reasoning, plan, complete code, trade-offs, edge cases, complexity, and validation.
  • Fast / Deep reasoning — concise fixes or thorough step-by-step analysis.
  • Your models, your keys — local LM Studio / Ollama, an existing Codex or Claude CLI login, or your own provider API key.
  • Encrypted local history — screenshots, prompts, and answers stay encrypted on your machine.
  • Screen-share aware — on supported Windows capture paths, Haired asks the OS to exclude its windows from Microsoft Teams, Zoom, and Google Meet shares.

Important

Use Haired only for content you are permitted to capture and send to your chosen provider. Screen-share protection is platform-dependent; always verify the live meeting preview before sharing sensitive content.

Features

Connect your own providers

No Haired account, no hosted proxy, no usage credits. Pick the model you already have.

Haired AI providers settings

Customize the overlay

Themes, opacity, fonts, shortcuts, magnifier, and launch-at-login—tune the overlay so it fits your workflow.

Haired appearance and behavior settings

Keyboard-first

Global shortcuts for capture, ask, settings, and moving answer windows. Record your own in the app.

Haired shortcut settings

Privacy check

See your platform's capture-protection status and run a local diagnostic before you share your screen.

Haired privacy check settings

Providers

Kind Providers
Local CLI (reuse existing login) OpenAI Codex CLI, Claude Code CLI
Local runtime LM Studio, Ollama
Bring-your-own-key OpenAI, Anthropic, Google Gemini, Mistral, OpenAI-compatible

API keys are encrypted with the OS secure storage (macOS Keychain / Windows DPAPI) and never returned to the renderer. Local models run on your machine; requests to BYOK providers go directly from the app to the provider.

Quick start

  1. Install Haired from the latest release.
  2. Pick a provider in Settings → AI providers:
    • Run codex login or claude auth login for CLI providers.
    • Start LM Studio's local server or ollama serve for local models.
    • Or paste an API key for OpenAI, Anthropic, Gemini, Mistral, or a compatible endpoint.
  3. Select a screen region with Cmd/Ctrl+Shift+Space (answer) or Cmd/Ctrl+Shift+Enter (ask).
  4. Read the answer in the overlay, copy code, ask follow-ups, or save it to encrypted history.
Action Default shortcut
Select & answer Cmd/Ctrl+Shift+Space
Select & ask Cmd/Ctrl+Shift+Enter
Open settings Cmd/Ctrl+Shift+H
Move answer window Hold Cmd/Ctrl+Alt + arrow keys

Preview builds are unsigned: Windows SmartScreen and macOS Gatekeeper may show a warning. Verify the SHA-256 checksums on the release page before installing.

License

MIT — see LICENSE.

About

Haired technical interview and coding copilot for Windows and macOS

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages